Practice Makes Perfect: Bundled Payment Models for 90-Day Total Knee Arthroplasty Episodes

Abstract

Commentary In this article, Schilling et al. provide a proper and elegant analysis of the variation in payments and resource utilization for total knee arthroplasty (TKA) immediately prior to the April 2016 implementation of the U.S. Comprehensive Care for Joint Replacement (CJR) bundled payment program, which was mandatory at the time of the study. Using the 100% sample of Medicare Part A claims data for this study, the authors included all elective unilateral primary TKAs performed in patients who were ≥65 years of age between January 1, 2014, and December 31, 2016. One methodological concern is a lack of description of the methods used to handle missing data, negative costs, and zero costs that exist with claims data. The inclusion of such data could impact the results, even with this large sample size of >700,000 TKAs. Regardless, this article contributes very good information to the body of knowledge of alternate payment models, and the authors even suggest that the risk-adjusted model could be adopted by hospitals to benchmark their performance in such models. However, questions remain about the degree to which the CJR program was responsible for the results and whether providers will successfully sustain these improvements. Schilling et al. illustrate that resource utilization, measured by Medicare Part A payments, length of hospital stay, readmissions, and inpatient post-acute care utilization, for Medicare patients undergoing TKA decreased nationally from 2014 to 2016. Therefore, these trends in decreased spending and resource utilization began prior to April 2016 when the CJR model was implemented. The CJR model was implemented in a time when other value-based initiatives were impacting TKA care redesign, particularly the Bundled Payments for Care Improvement (BPCI) program, which began in 2013 and broadly influenced utilization rates in participating markets1,2. Furthermore, the early results of BPCI Advanced suggest that retrospective shared savings models such as CJR and BPCI create a floor in market pricing, after which incentives become too low to elicit participation1-3. Although the CJR model participation was initially mandatory, voluntary participation continues through December 2020. Early evaluations have suggested that the CJR model was somewhat successful. Barnett et al. reported a modest 3% decrease in spending with no significant changes in postoperative complications4. It is reasonable to posit that the gradual transition from a fee-for-service payment model to a value-based care model impacted the decrease that Schilling et al. report. Although payment reform has been underway since 1983, the first bundled payment model for orthopaedics began in 2009 with the Acute Care Episode (ACE) program5. The testing of this model was narrow in scope, with just 5 hospitals participating5. In 2013, the Medicare BPCI initiative began, with >1,200 participating entities in July 20183. The BPCI initiative substantially broadened the scope to include physician groups and post-acute facilities, in addition to hospitals. Most participants chose a retrospective shared savings model that focused on reducing post-acute utilization, which drove down the mean episode costs in participating markets, without degradation in outcomes. This had 2 important effects: (1) early participants developed best practices for the BPCI program that the CJR program participants were able to replicate to perform well with respect to all shared savings, and (2) strong performance in the BPCI program lowered the mean target spending for Medicare nationally for this procedure, thus reducing the opportunity for further shared savings in subsequent programs in a given market. There was an initial 16% decrease in participants just before the March 1, 2019, deadline to enroll in the BPCI Advanced program6. Presumably, this is because institutions and providers cannot lower costs and readmissions sufficiently more to warrant participation. The time periods of the BPCI and CJR programs overlapped and both included Medicare patients2,3. Therefore, the trend in the improved value of TKA illustrated by Schilling et al. includes the successful impacts of the BPCI program. The CJR program began in April 2016 and, at the time, it was mandatory for total joint replacement unless providers were participating in the BPCI program, which began in 20132,3. The design of the CJR bundled payment model was informed by the BPCI bundled payment model; however, physician groups could be episode initiators with the BPCI program but not with the CJR program. Regardless, the effects of the BPCI model are present in the data used in this study. This does not negate the importance and utility of this study by Schilling et al. to serve as a benchmark for future bundled payment models that will include procedure-based bundles.